Is reversible mapping between FHIR and Archetypes possible?

Question

I'm trying to understand the risks involved in one system 'speaking FHIR' and another 'speaking OpenEHR/Archetypes'. Especially for the Observation, Questionnaire, Device and Patient resources.

Can the two content formats be mapped between each other seamlessly, possibly utilizing some extensions where required? What are the major issues in such a mapping, and are there people/organisations/vendors who have done such mappings?

Answer 1

The two formats can be mapped to one another, but possibly not seemlessly. They differ fundamentally in scope and intent. And there's different issues for each resource you ask about

• openEHR doesn't have a fixed Patient. The standard patient archetype on the openEHR CKM seems very perfunctory, and includes things FHIR puts in other resources (e.g. Coverage), and has no coverage I can see for important concepts in FHIR like active.

• I'm not really sure where device even manifests in the openEHR world. Perhaps the agent resource?

• Questionnaire doesn't have an openEHR equivalent directly, except maybe an templated archetype is pretty much a questionnaire, though I don't think that the intent is to have a dizzying myriad of templated archetypes the way that FHIR expects there to be questionnaires of no particular fixed semantics

• Observation - openEHR doesn't really have a base Observation, except as defined in the reference model. That base observation is kind of compatible with the FHIR Observation but the real work is in the Observation archetypes, and there's lots of them. Many make variable assumptions about the underlying reference model, and about how observations work. And the base FHIR observation doesn't handle things that the openEHR reference model does, such as time series.

I know of no organizations that have done systematic mappings of the particular resources that you mention. There is some work afoot; for instance, we co-developed the AllergyIntolerance resource and the AdverseReactionRisk archetype, though I don't know how sustainable that will prove. It was hard work to align the community processes and analyses, and it might not be repeatable

Fundamentally, FHIR and openEHR are doing different things, and have different strengths and weaknesses. A system implementing both is going to do manual mapping between it's internals and externals. Perhaps that will change slowly in the future, but not in any hurry

Answer 2

Mappings can be done but is case by case, there is't a set of rules that can be automated for any FHIR resource and any openEHR archetype.

Also you can create archetypes adjusted to FHIR resources, or profile FHIR resources to match existing openEHR archetypes.